Electronic Distributor Adapter

[90TurboVan]

Rewrote the whole thing, and it works on the bench. Being as I'm testing it on exactly what I scoped, it should work fine when I try it out on my Neon, which should happen tomorrow. If all goes well, your adapter will be out Friday, turbo.shaft.

The zip has the new code, which is fully commented, not like the old code...

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan Rewrote the whole thing, and it works on the bench. Being as I'm testing it on exactly what I scoped, it should work fine when I try it out on my Neon, which should happen tomorrow. If all goes well, your adapter will be out Friday, turbo.shaft. The zip has the new code, which is fully commented, not like the old code...

If this works I will be whistlin' Dixie from all my orifices

Looks like it may be going back into the Woody wagon after all.

This is cutting it close to the end of the car season for me. I start as a Killington Ambassador this Saturday for Skiing on the 16th so I barely have weekend debugging time left at my friends project garage on weekends.

[90TurboVan]

You can whistle from one at least. It synchs and outputs properly, but as soon as you tach it up a bit it loses synch until the removal and reapplication of power. I guess I'll have to figure a way to add rpm control into my simulator.

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan You can whistle from one at least. It synchs and outputs properly, but as soon as you tach it up a bit it loses synch until the removal and reapplication of power. I guess I'll have to figure a way to add rpm control into my simulator.

You are simulating the point to trigger based on predicted speed. Is it a fast enough system to recheck its speed estimate after each rotation of the engine? It seems as if it is not a dynamic enough system, which is even more important with a turbo.

There is not this problem with the 2003+ NCG 2.0/2.4 motors also? or is it?

I could modify my head and put it on a cheap srt shortblock or just get a new head.

[90TurboVan]

Oh, its definitely fast enough. The PIC runs faster internally than the 68HC11 in the SMEC/LM does, and it has less to do. I'm pretty sure I just didn't handle some of the subratction/division calculations correctly since they were extended beyone the native 8 bit capabilities of the PIC. I haven't actually hooked one up to an SRT yet, but the NGC stuff is all hard triggered from the teeth on the crank, since they're evenly spaced, so it shouldn't have any problems at all.

It also does not calculate crank rpm, it takes the time between two crank falling edges (20 crank degrees), divides it by two, and counts off that time to get a simulated pulse at every 10 degrees. It measures the change in time between two consecutive 20 degree pulses and uses that to predict how the 10 degree time will change. If that isn't done right, it could lose synch which is what I think its doing.

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan Oh, its definitely fast enough. The PIC runs faster internally than the 68HC11 in the SMEC/LM does, and it has less to do. I'm pretty sure I just didn't handle some of the subratction/division calculations correctly since they were extended beyone the native 8 bit capabilities of the PIC. I haven't actually hooked one up to an SRT yet, but the NGC stuff is all hard triggered from the teeth on the crank, since they're evenly spaced, so it shouldn't have any problems at all. It also does not calculate crank rpm, it takes the time between two crank falling edges (20 crank degrees), divides it by two, and counts off that time to get a simulated pulse at every 10 degrees. It measures the change in time between two consecutive 20 degree pulses and uses that to predict how the 10 degree time will change. If that isn't done right, it could lose synch which is what I think its doing.

Is this calculation done for every revolution?

How often is it updating to a fixed/measured point to correct for drift/error/etc.

If I understand this correctly the batch firing is done approximately at 720 and 1440 degrees rotation of the crank before a new firing cycle begins. Is this correct? Where is the reset point for timing or is there not one?

Without a reset point carried error builds till there is a huge error. Thus if timing was lost during revving it meant that many cycles had passed without a hard timing reset allowing for time drift caused by carried error.

Is this a correct assumtion?

[90TurboVan]

You assume correctly, however, it synchs every TDC. There are 4 falling edges on the crank sensor before each TDC, they are each 20 degrees apart. It measures all three, subtracts the first one and middle one to get one delta, subtracts the middle one and the last to get the second delta, and subtracts the deltas to get the second derivative. Then they're all divided by two to get the values for each 10 degrees. At each final TDC falling edge (actually 9 degrees before), it forces the HEP outputs low and recalculates all the deltas. Subsequent 10 degree increments are calculated form that "hard" falling edge untill the next crank pulses, when the whole process starts over again. It also stops counting off increments after 180 degrees, so it can't "run away" so to speak. I also caught that before you edited, and I said "calculates rpm", which it doesn't do

It actually had nothing to do with how I handled the deltas... I didn't have some variables initializing to a defined point at the start, so stuff got a little out of hand once it start using those values. Fixed/improved some other minor things that I don't remember and made my 2.4 simulator take an rpm input. I don't see variation on the scope when I crank the rpm up fast, so everything seems to be working properly. I'll put it on my Neon once again tomorrow and see what it does.

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan You assume correctly, however, it synchs every TDC. There are 4 falling edges on the crank sensor before each TDC, they are each 20 degrees apart. It measures all three, subtracts the first one and middle one to get one delta, subtracts the middle one and the last to get the second delta, and subtracts the deltas to get the second derivative. Then they're all divided by two to get the values for each 10 degrees. At each final TDC falling edge (actually 9 degrees before), it forces the HEP outputs low and recalculates all the deltas. Subsequent 10 degree increments are calculated form that "hard" falling edge untill the next crank pulses, when the whole process starts over again. It also stops counting off increments after 180 degrees, so it can't "run away" so to speak. I also caught that before you edited, and I said "calculates rpm", which it doesn't do It actually had nothing to do with how I handled the deltas... I didn't have some variables initializing to a defined point at the start, so stuff got a little out of hand once it start using those values. Fixed/improved some other minor things that I don't remember and made my 2.4 simulator take an rpm input. I don't see variation on the scope when I crank the rpm up fast, so everything seems to be working properly. I'll put it on my Neon once again tomorrow and see what it does.

So you weren't doing a hard update point every rotation until now? I am confused. What version was running away and required removal of the throttle to reset? What were you doing then?

How is it running now

[90TurboVan]

They all did hard points, it just seemed to lock up and need to be reset when you'd tach it up. Traced most of the problems to how the outputs are handled, some jump offsets make the code run away. Just got back from testing it out once again, still have problems with lost synch. Never at idle, only at higher rpm's, and nothing that I can reproduce on my simulator. I'm going through the code right now to look for places that it can get stuck or run away. The problem isn't with how the adapter works, its something simple that I'm missing. Sometimes I hate writing in assembly, all the little stuff like this is handled by the compiler in C.

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan They all did hard points, it just seemed to lock up and need to be reset when you'd tach it up. Traced most of the problems to how the outputs are handled, some jump offsets make the code run away. Just got back from testing it out once again, still have problems with lost synch. Never at idle, only at higher rpm's, and nothing that I can reproduce on my simulator. I'm going through the code right now to look for places that it can get stuck or run away. The problem isn't with how the adapter works, its something simple that I'm missing. Sometimes I hate writing in assembly, all the little stuff like this is handled by the compiler in C.

Good luck. This is why I don't have the patience to program. Have you tried doing the NGC stuff yet? Does it have the same bugs? Are the bugs all related to the lack of coinciding hard points?

[turbo.shaft]

Would a 2.4 from a 2003 caravan have a NGC motor? or is it only the SRT4?

[90TurboVan]

2003+ should be NGC. If it has the crank sensor on the front of the motor right below the starter, its NGC. My programmer decided to stop working on me, which is no fun at all. I'm off of school most of next week, so I'll actually have time to work.

[turbo.shaft]

Quote:

Originally Posted by 90TurboVan 2003+ should be NGC. If it has the crank sensor on the front of the motor right below the starter, its NGC. My programmer decided to stop working on me, which is no fun at all. I'm off of school most of next week, so I'll actually have time to work.

So how much longer for a running prototype?

Would it be quicker to get an NCG car electronics running, because I can make that happen if you can get me a workign prototype. I have an SRT shortblock, just need to hunt down a matching head.

[90TurboVan]

Like I said, I have off the rest of this week. The adapter will be my main focus/project. I'm all set up for testing non-NGC right now, and I'd really like to get it to work, I also don't want to make you go through changing a motor. Got my programmer working again this past weekend, so I'll start testing again tonight.

[nolydi86conques]

ok well i would like one fore my 1985 conquest 2.6 pm me if u can help